The purpose of this study was to evaluate the effects of a 6-session yoga
program during pregnancy on maternal comfort, labor pain, and birth outcomes of
primiparous women. In this chapter, the methodological aspects of research design,
setting, population and sample, instruments, experimental intervention, protection of
human subjects, and data management are presented.
A randomize controlled trial with an experimental of repeated measures design
was used to investigate the effects of the yoga program during pregnancy on maternal
comfort, labor pain, and birth outcomes. Experimental control consisted of the
inclusion and exclusion criteria, randomized minimization techniques to control for the
confounding variables in the study.
This study took place in the antenatal clinics in two settings at public hospitals
in Southern Thailand. These hospitals provide similar standardized care. Currently, the
maternity service in Thailand requires all pregnant women to have at least four
antenatal care visits (Ministry of Public Health, 2002). However, the number of
antenatal visits varies depending on the women’s condition and consideration by the
health-care provider. A common recommendation will be that a pregnant woman
should use the antenatal service when she recognizes her pregnancy. The woman
attends antenatal check-up every month until 28 weeks of gestation, then every two
weeks from 28-36 weeks and every week after 36 weeks.
Antenatal care is provided free of charge by public-health services. Generally,
obstetric-gynecological nurses or nurse-midwives provide this care in a normal
pregnancy, and an obstetrician gives care to a pregnant woman at least once during
antenatal care. An obstetrician or gynecologist is responsible for women with a high-
risk pregnancy. Women may, however, choose to have their own private obstetricians,
and hence have antennal check-up in their doctor’s private clinics or in the hospital
where the doctor is attached as a medical specialist. This can be at a public or a private
hospital. Most births occur in hospital or health-care settings. In general, a woman
gives birth where she receives her antenatal care. Where antenatal care is received
from a private doctor, the birth will take place in a hospital where the doctor works as
a medical specialist (Liamputtong, Yimyam, Parisunyakul, Baosoung, & Sansiriphan,
Two sub-settings at each hospital were involved in this study, the antenatal
clinic and the labor room. The antenatal clinic provides the fundamental services
during prenatal visits including: estimation of gestational age, weight monitoring,
screening for sexual transmitted diseases, immunization, screening for maternal
complications, and evaluation of fetal well-being. Nurses in the antenatal clinic usually
educate pregnant women about diet, rest, hygiene, and identifying the immediate
symptoms that they should seek treatment for from the hospital. These are bleeding
from the vagina, a decrease or absence of fetal movement, a headache, or other related
symptoms. Minor ailments during pregnancy and their management are also included,
such as morning sickness, leg edema, and backache.
During the delivery period, pregnant women are admitted if there is evidence of
true labor or certain conditions such as ruptured membranes, placenta previa or
abruptio placenta. Nurses assess, monitor, and document the data of both the mother
and her baby. Without precaution, a small amount (50cc) of unison enema and delivery
preparation may be given. Relatives are not allowed to give support to the women after
admission. To monitor the labor progression, the assessment of uterine contraction
with the fetal heart sounds every 30-60 minutes and vaginal checking approximately
every two hours are performed.
Population and Sample
The population in this study consisted of primiparous women who attended the
antenatal clinics and delivery rooms of the target settings from January 2005 to
February 2006. Pregnant women who met the inclusion criteria were approached and
informed of the study. Written informed consent was obtained from those who verbally
agreed to participate. In order to reduce bias from systematic selection as it balanced
potential confounding (prognostic) variables, and equalized the number of subjects in
the two groups (Zeller, Good, Anderson, Zeller, 1997). They were randomly assigned
by using the minimized randomization program version 2.01 (Zeller, 1997) (using
Microsoft Access 2000) to either the yoga or control group. This method was used to
take into account maternal age, education, income, marital status, and trait anxiety.
The inclusion criteria were as follows: (1) primiparous women, to control for
effect of past labor pain, of at least 18 years of age (in accordance with Thai law for
signing an informed consent form): (2) ability to communicate and write in the Thai
language because this would have an effect on the reliability of responses to the
questionnaire; (3) no serious illness or high risk complications during pregnancy (such
as pregnancy induced hypertension and bleeding from the vagina) that could be easily
threatened by yoga practice; (4) no intention to give birth by cesarean section or
painless labor; (5) attending antenatal care from the beginning or at least from the
second trimester of pregnancy; (6) agreeing to be involved in either group; and (7)
having no prior experience of yoga practice.
Pregnant women were excluded from the study if they (1) received induction of
labor, or a caesarean section; (2) had complications during the study such as pregnancy
induced hypertension, bleeding from the vagina, or had complications during labor and
after giving birth, and (3) were unwilling to participate at any time during the study.
A previous study, following a similar process of a randomized, single blind,
controlled trial, examined the effects of an intervention using supervised yoga and
relaxation techniques specially designed for patients with carpal tunnel syndrome on
pain intensity (Garfinkel et al., 1998). The yoga-based intervention group took part in a
program which used yoga techniques for strengthening, stretching, and balancing each
joint in the upper body along with relaxation to help counteract the energy-draining
effects of prolonged stress and chronic pain. Therefore, it was used as an example for
sample size calculation. The effect size was calculated on the basis of the mean and
standard deviation taken from that study. The mean of pain intensity of VAS of the
experimental group ( x = 5, SD = 2.8) was significantly lower than that of the control
group ( x = 2.9, SD = 2.2) at p = .02. The effect size (d) calculated for pain intensity
was 0.64. To achieve a power of .80 at alpha .05, an adequate approximation of n is
given by the following formula (Cohen, 1988, p. 53):
n = n.10 +1 = 1571 +1
100d2 100 (0.64)2
Cohen’s (1988) Table 2.4.1 on page 55 was used in calculating the necessary
sample size value for the given significant criterion of a2 = .05, which n.10 = 1571 and
desire power at effect size (d) = .10 indicating a minimum requirement of 39 subjects
per group. The total sample in the study was 78.
There were only 74 primiparous women enrolled in this study, of which 66
cases were completed in the delivery period (Figure 6). The causes of dropout were
incomplete data due to the active progression of labor as 6-8cm of cervical dilatation
on admission, and using a cesarean section due to cephalopelvic disproportion before
the delivery period. The reasons for dropping out of the control group were
unwillingness to participate any more, precipitate labor, and having a cesarean section
due to placenta previa and abruptio placenta before the 1st point of data collection.
Thus, the attrition rate in both groups in the current study was 10.81%.
Subjects enrolled in Hospital A Subjects enrolled in Hospital B
Experimental Group Control Group Experimental Group Control Group
(n = 2) (n = 2) (n = 35) (n = 35)
Subjects enrolled in the study (n = 74)
Allocated to Experimental Group Allocated to Control Group
(n = 37) (n = 37)
(n = 4) (n = 4)
Assessed and analyzed Assessed and analyzed
during delivery period during delivery period
(n = 33) (n = 33)
Figure 4 Flow chart of the sample population
The instruments in this study were classified into two categories as follows:
1. An intervention: The yoga program. It was developed based on yoga principles
(Balaskas, 2003; Gore, 1997; Kuvalayananda & Vinekar, 1994; Udupa, 1985;
Yogendra & Desai, 1994; Yogendra, 1998) and combined with the researcher
experience of engagement in yoga training. To evaluate the feasibility of the program
intervention, and the study instruments. A pilot study of 3 participants who had the
same criteria as in the study was carried out.
In order to ensure the stability of the experimental intervention, the yoga
booklet and a cassette were provided for the subjects and used as a reference with
instructions to follow. Details of the booklet and the cassette are as follows.
1.1 The booklet for yoga practice consists of information regarding preparation
for childbirth and pain management that encompass yogic principles. Steps and
benefits of each yoga asana, pranayama, and deep relaxation technique are included
1.2 The yoga cassette tape guided the yoga practice and combined a sequence
of asana, pranayama, yoga nidra, and meditation. This cassette tape provides guidance
modified from the program yoga for health, which was proposed by Kasetsomboon
(2001) for approximately 60-70 minutes.
2. Instruments used for data collection: Demographic Data Form, Trait Anxiety
Form, Maternal Comfort During Pregnancy (MCDP), Maternal Comfort During Labor
(MCDL), Visual Analogue Scale to Total Comfort (VASTC), Visual Analogue
Sensation of Pain Scale (VASPS), Pain Behavioral Observation Scale (PBOS), and
Yoga Practice Record Form.
2.1 The Demographic Data Form (Appendix A) consisted of maternal
information including; age, religion, education, marital status, occupation, income,
gestational age, serious illnesses or complications during this pregnancy, types of
delivery, pain medication, serious illness or complications during and after delivery,
and duration of birth and delivery. The newborn’s data consisted of gender, gestational
age, birth weight and Apgar score at one and five minutes after birth, serious illness or
complications and illness that required transferring the newborn to a neonatal intensive
care unit. This instrument was developed in two parts; the first part was the
demographic data form that pregnant women filled in by themselves after signing the
consent forms and before randomized assignment to groups. The second part was the
data that was obtained from the mother’s records and her infant’s birth records by a
research assistant in the delivery period.
2.2 The State Trait-Anxiety Inventory, Trait Anxiety Subscale (Appendix B)
was used. This is one of the most widely known self-report scales used to measure
anxiety, and which has been used extensively as a research and clinical instrument, and
originally developed by Spielberger (1983). This scale contained statements for the
respondent to choose from to describe how she generally felt, and this was
conceptualized as a relatively stable individual difference in proneness to anxiety.
Validation of this scale had proved its reliability and sensitivity in measuring anxiety.
The trait scale contained 20 items using a 4-point Likert scale; each item could be
answered on 1-4 scale, where ‘1’ stood for ‘almost never’ and ‘4’ stood for ‘almost
always’. The total gave scores of 20-80. A breakdown was made of baseline score
categories between 20-39, 40-59, and 60-80, which indicated low, moderate, and high
trait anxiety scores respectively (Sesti, 2000). Trait anxiety, as a controlled variable,
was measured only at the beginning of the study.
2.3 Maternal Comfort During Pregnancy (MCDP) (Appendix C) and Maternal
Comfort During Labor (MCDL) (Appendix D) were instruments modified from The
General Comfort Questionnaire (Kolcaba, 1992). To do this, positive and negative
items were generated for each cell in a two-dimensional grid of the three states of
relief, ease, and transcendence, with the four contexts of physical, psychospiritual,
social, and environment, in accord with the theory of holistic care. Twenty-four
positive items and eleven negative items were compiled, using a Likert-type format
that ranged from strongly agree to strongly disagree. Responses were scored from 1 to
6; the higher score indicated more comfort. In this study, it was used with both groups
to measure maternal comfort experience at the beginning of the study (26th-28th),
34th, 37th week of gestation, and at the second hour after the delivery.
2.4 Visual Analogue Scale to Total Comfort (VASTC) (Appendix E) was the
instrument used to assess maternal comfort of childbirth experience during pregnancy
and during labor. This VASTC identified relief, ease, and transcendence to signify
factors derived from the previous factor analysis of the General Comfort Questionnaire
(Kolcaba, 1992), and total comfort. Stems used for VASTC were: total comfort - “I
feel as comfortable as possible right now”; relief - “I have many discomforts right
now”; ease – “I am feeling at ease right now”, and transcendence - “I am feeling
motivated, determined, and strengthened right now”. Responding to the scales required
each participant to place a mark on each 100mm horizontal line that had anchors from
the left end of the line in millimeters for scoring total comfort. The current study used
only the total comfort item at each yoga session during antepartum period, and at the
same time as measuring the labor pain.
2.5 The Visual Analogue Sensation of Pain Scale (VASPS) (Appendix F) was
used to measure the labor pain by self-reporting on a horizontal line. This VAS
consists of a 100mm. line anchored by two extremes of pain, “no pain” and “worst
imaginable pain”. Severe pain was defined as a mark over 60mm on the VAS,
moderate pain as a mark from 40-60mm, and mild pain as a mark under 40mm on the
VAS (Maxwell, 1978; Revill, Robinson, Rosen, & Hogg, 1976). Each parturient was
asked to put a mark through the line corresponding to the intensity of pain from uterine
contractions. The scoring was measured from the distance in mm. from “no pain” to
the patients’ marks.
According to Kendrick and Simpson (2001) the active phase of labor averages
4.6 hours, and uterine contractions every 2-5 minutes in regular frequency, are 40-60
seconds in duration, and have moderate to strong intensity by palpation. Maternal
physical sensations are increasing discomfort; trembling of thighs and legs, pressure on
bladder and rectum, and backache. Maternal behavior consists of: beginning to work at
maintaining control during contractions; becoming quieter; and accepting “coaching”
efforts of perinatal staff and support persons. There are tremendous changes in the
transition phase, with uterine contractions every 1.5-2 minutes of 60-90 seconds
duration, and the mother feels the urge to push, and increase rectal pressure. Maternal
behavior includes; ambulating; finding difficulty with uterine contractions; being
irritable and agitated; being self absorbed and may appear to be sleeping between
contractions; asking for more support; verbalizing feelings of discouragement and
doubting their ability to cope. Therefore, in this study, VASPS was selected to measure
at the starting point of the data collecting procedure, when the cervical dilatation was
3-4cm. and uterine contractions were of 30-60 seconds duration, and also at the end of
each two hours twice in the active phase of labor as presented in Table 1. To prevent
response bias, the women were not allowed to see the scales they had marked
previously (Phumdoung, 2003).
2.6 The Pain Behavioral Observation Scale (PBOS) was used to record the
parturient behavior in pain assessment. This instrument based on the Observation
Checklist of Laboring Women’s Behavior by Boasoung (1983) (Appendix G). The
behavioral checklist consists of five aspects in the observation of vocalization
expressed, body movement, breathing control, facial expression, and communication.
Each part used a 3-point Likert scale of 3, 2, and 1 scores respectively. It gave scores
of 5-15, whereby the low score indicated severe pain and the high score indicated mild
pain on the part of the parturient. The assessment was made by the research assistant
who recorded the behavior expressed by laboring women at the same time as
measuring the VASPS. It was measured from the beginning of the uterine contraction
until uterine relaxation, and the results were recorded immediately.
2.7 The Yoga Practice Record Form (Appendix H) is a self-report measuring the
length of participation by the experimental group engaged in the practice of the Yoga
program techniques during the previous week. Using the Self-Help Intervention
Project (SHIP) (Braden, Mishel, Longman, & Burn, 1989) as a guide, the course was
organized into six class sessions. The participation actually undertaken by the subjects
assigned to the self-help course experimental group was quantified according to the
number of class sessions they attended. A record of the subjects’ attendance at the
sessions served as the instrument for measuring the exposure received. A continuous
scoring scheme was used to represent the exposure variable in the subsequent analysis.
Participants in the control group and those in the experimental group who did not
attend any of the classes were assigned a value of 0; for the other participants, they
were assigned values ranging from 1 to 6, based on the actual number of classes they
In the current study, participants rated how often they practiced and the number
of minutes each time. The target was set at 30-60 minutes of practice a day for at least
three days a week. This record form was kept in order to determine whether the effect
of yoga practice on labor pain, maternal comfort, and birth outcomes would be affected
by exposure to yoga practice (duration and frequency). The exposure to yoga practice
was measured by the frequency and duration of practicing. The scoring was as follows:
if the frequency of yoga practice of 30-60 minutes per day was less than three times
per week, the assigned score was 0; if it was 30-60 minutes practice per day three
times per week the score was 1; if it was 30-60 minutes practice per day four to five
times per week it was 2, and if it was 30-60 minutes practice per day six to seven times
per week it was 3. If the duration of yoga practice was 30-45 minutes each time, the
assigned score was 1; if it was 46-60 minutes each time it was 2; and if it was 61-75
minutes each time it was 3.
The instruments used for collecting the data in this study were: the
Demographic Data Form, the Intrapartum Chart Abstraction Form, the State Trait
Anxiety Inventory: Trait Anxiety Form, the Maternal Comfort During Pregnancy Form
(MCDP), the Maternal Comfort During Labor Form (MCDL), the Visual Analogue
Scale to Total Comfort (VASTC), the Visual Analogue Sensation of Pain Scale
(VASPS), the Pain Behavioral Observation Scale (PBOS), and the Yoga Practice
Record Form. These self-report instruments, excluding the Demographic Data Form,
PBOS, and Yoga Practice Record Form, were originally developed in the English
Language. In order to ensure culturally equivalent versions of these instruments in the
Thai language, a translating process using decentering and back translation technique
(Brislin, 1980) was conducted. The modified Thai version was judged by a panel of
Thai experts to ensure cultural validity. These experts were also asked to rate
independently the degree of cultural relevance of each item of the instruments. The
results of the ratings were used to compute the cultural validity index (CVI), using the
same process as computing the content validity index described by Waltz, Strickland,
and Lenz (1991). The CVI of MCDP was 0.90, and MCDL was 0.95.
Thai versions of the Demographic Data Form, State Trait Anxiety Inventory:
Trait Anxiety Form, MCDP, MCDL, VASTC, VASPS, PBOS, and Yoga Practice
Record Form were assessed for validity of content by five experts. These were an
obstetrician and four nurse instructors with expertise in the area of obstetric care or
who have had experience in using these or relevant instruments in their research work.
Suggestions from the experts were incorporated in the final revision of the
questionnaires, visual analogue scales, and the pain behavioral observation scale. The
booklet and the yoga cassette tape guide had been sent to two experts who had
experience in yoga training and an obstetrician to assess content validity. In order to
check the clarity and appropriateness of the booklet and the yoga cassette tape guide
they were piloted with three pregnant women who satisfied the inclusion criterion for
1. The maternal comfort questionnaires and trait anxiety subscale were tested
with primiparous women who met the inclusion criteria. Cronbach’s alpha coefficients
were computed to ascertain internal consistency reliability of the trait anxiety and
maternal comfort questionnaire. The internal consistency estimate of the translated
version of the MCDP was 0.89, and the MCDL was 0.86, which were acceptable for
using the newly translated instruments in a different culture. The internal consistency
of the State Trait Anxiety Inventory: Trait Anxiety Scale was 0.85.
2. The Visual analogue sensation of pain scale (VASPS) was tested for
reliability by using the test-retest method. In this study, women were asked at 3-4cm of
cervical dilatation and when uterine contractions were of 30-60 seconds duration, and
also at the end of two hours after the first time of the active phase of labor. The internal
consistency of the VASPS was 0.74. The visual analogue scale to total comfort
(VASTC) was also tested for reliability by using the test-retest method at two-week
intervals during pregnancy and at the same time of testing VASPS during labor. The
internal consistency of VASTC during pregnancy was 0.91 and during labor was 0.80.
3. The pain behavioral observation scale was tested for internal consistency by
using the interraters observed and recorded parturient behavior of the same checklist.
The agreement between the interraters, the reliability coefficient, was determined by
the following formula (Selltiz, Wrightsman, & Cook, 1976):
Reliability coefficient = Number of Agreements
Number of Agreements + Number of Disagreements
In the current study, the value of agreement of PBOS was 0.80.
The data-collecting plan is shown in Table 1.
Data Collecting Plan
Antenatal Period (Week) Delivery Period
26-28 30 32 34 36 37 C x- Dilated 2 hrs after 2 hrs after Two hour
Time 3-4cm Time 1 Time 2 after birth
1. Demographic Data Form √
2. Trait Anxiety Form √
3. Maternal Comfort During √ √ √
4. Maternal Comfort During
Labor (MCDL) √
5. Visual Analogue Scale √ √ √ √ √ √ √ √ √
to Total Comfort (VASTC)
6. Visual Analogue Sensation √ √ √
of Pain Scale (VASPS)
7. Pain Behavioral Observation √ √ √
8. Yoga Practice Record Form √ √ √ √ √ √
Length of Labor √
Apgar Score √
The experimental group received a series of six 60-minute practice sessions of
the yoga program at the 26-28th, 30th, 32nd, 34th, 36th, and 37th week of gestation
respectively, whereas the control group received standard care. Table 2 shows the yoga
program component and the descriptions of the guidelines to be practiced. The yoga
program was a combination of two elements: educational activities giving a brief
description of basic anatomical structures related to pregnancy and yoga explaining the
concepts related to each component. Yoga asanas, pranayama, yoga nidra, dhyana,
and a combination of chanting mantra were practiced harmoniously and in an orderly
manner to improve flexibility, muscle force control, endurance, balance and body
coordination of the body segment, to reduce fatigue, to relieve some of the discomfort
of pregnancy, to promote body awareness and concentration, to elicit a relaxed
response, and to promote mental clarity by focusing on the sound of breathing.
Additionally, the participants were provided with a booklet and tape cassette for self-
study use, which explained the principles and benefits of each yoga practice and asked
the mothers to practice at home at least three times a week starting after the first yoga
session and continuing until their sixth session. It was expected that each subject
would practice yoga for a period of 10-12 weeks. Subjects were informed that they
could practice more than three times a week. Compliance in both groups was ensured
through the research team making frequent telephone calls, and an additional request
was made for subjects in the experimental group to maintain a strict record in a diary.
Only the researcher was responsible for the yoga program. All routine hospital
procedures were done similarly to the procedures used for the control group. Subjects
in the control group received routine hospital procedures and standard nursing care by
the staff nurses, and the researcher engaged them in casual conversation for 20-30
minutes during each session. Table 3 represents the Yoga program session objective as
a guideline to practice.
Yoga program practice components
Education (5-10 min.) Subjects were given a brief description of basic
anatomical structures that related to pregnancy
and birth. The goal was to facilitate a greater
understanding of the physical function during
pregnancy and labor.
Body awareness (5-10 min.) Subjects were guided through awareness of
various parts of their body and their thoughts. The
aims were to promote awareness of body
sensation, position, and awareness of the activity
of the mind.
Pranayama (breathing exercise) Voluntary breathing activities were taught and
(5-10min.) practiced, such as, a complete breath of 3 parts.
The goals were to promote awareness of the
sensation of the breath in the body, and awareness
of how the breath could facilitate movement of
body segments, and to promote concentration.
Table 2 (continued)
Yoga program practice components
Asana (physical poses) and Subjects were instructed and assisted as necessary
Pranayama (breathing exercise) in performing a variety of modified yoga poses
(15-20 min.) related to each Yoga program session objective.
The goals were to improve flexibility, muscle
force, endurance, balance, and coordination of the
body segments. It is also restored to the body-
mind, its fundamental condition of the well-
being, ease, and vibrant alertness.
Yoga Nidra (relaxation) The subjects were instructed to listen and follow
(5-10 min.) guide imagery tape incorporation visualization
and then allowed to rest in silence. The goal was
to elicit a relaxation response.
Meditation (5-10 min.) The subjects were instructed to be silent, focusing
on the sound of breath while sitting or lying
down. The goal is to promote mental clarity (clear
one’s mind of extraneous thoughts).
Yoga program sessions
1. Establishment of a solid foundation The educational purpose was focused on
information about a brief description of
basic anatomical structures related to
pregnancy and birth, everyday posture
harmony with the pregnant state, and
warm up exercises.
2. Promoting comfort The focus of this session was on the
asanas for releasing tension.
3. Activating the power of birth passage The focus was on essential exercises,
especially on the pelvis, in order to
prepare the baby for the birth.
4. Knowing the flow of life and birth It focused on the pranayama or breathing
techniques and visualization for birth.
5. Creating better balance It emphasized the stability of mind-body
6. Creating peace of mind This session focused on the feeling of
Preparation for the Researcher and Research Assistant
The researcher had been engaged in yoga training at the Yoga Club, Faculty of
Nursing, Prince of Songkla University. In addition, the researcher practiced yoga
following the tape cassette guide at least two times in order to gain experience and skill
in delivering the yoga program.
There was one research assistant who had a bachelor’s degree in education
assisted in conducting the Yoga program for pregnant women, and also collected data
in the antenatal period. Before the data collection, a research assistant engaged in yoga
training with experts at the Yoga Club of the Faculty of Nursing, Prince of Songkla
University at least two times to attain a good standard of yoga practice.
The other research assistants were the nurses who had at least one year’s
experience in taking care of laboring women. The researcher trained all the research
assistants who helped in both the antenatal and the delivery periods in all aspects of
data collecting. The instructions for recruiting potential participants, administering the
instruments, and issues pertaining to informed consent and the use of human subjects,
were all raised and discussed. Furthermore, the researcher supervised each research
assistant in collecting data from the first few subjects. Practical problems, and a written
document related to collecting procedures and relevant issues, were discussed and
resolved and suggestions were noted for the further collecting of data.
This study took place in two settings in antenatal and labor units at two public
hospitals. There was a study data collector manual available in which contained
information essential to standardized data collection. In addition, antenatal clinic staff
and data collectors working as a team in the labor room from each hospital attended an
orientation providing an overview of the study before the program began. This process
included a discussion of informed consent; confidentiality and data collection issues;
and a discussion of the key concepts and variables used in each of the data collection
tools. Each tool was included in the study manual, and oral presentations of the
content, with specific examples of how to record the data correctly, were given in the
preparation phase. Timing of delivery and birth could not be exactly predicted,
whether during the morning, evening or night shifts. However, the primary contact
person and the data collecting nurses had the opportunity to meet and get to know one
another at the orientation session, which was also used for planning in connection with
the continuing collection of data from subjects. They were reminded of the importance
of their participation as data collectors in this study. After the study finished, monetary
incentives were paid (100 Baht per case) to the data collectors and a letter of thanks
When the research proposal had been approved, the researcher met the nurses
at the antenatal clinics to explain the purpose and the nature of the study. Potential
participants were accessed through the registration lists at the antenatal clinics. They
were approached and informed about the investigation of the effects of the yoga
program during pregnancy, and during labor and delivery. They were invited to
become involved in the study by assessing the response to the experimental
intervention at each scheduled appointment and during the labor period. They were
assured that they would receive the same standard care from the hospital as those who
did not participate in the study, and they would receive a monetary incentive of 200
Baht for participating in the program. Both groups would receive a letter of thanks for
their participation. After verbal agreement to participate was obtained, the informed
consent form (Appendix I) was read and explained to the subjects. The informed
consent form included information regarding the study’s purposes, time commitment,
procedure to be used, and the right to withdraw at any time without affecting the status
or the care they received.
Protection of Human Subjects
In relation to the rights of human subjects, the Ethical Research Committee,
Faculty of Nursing, Prince of Songkla University, and the hospital directors and the
hospital ethical committees approved the research proposal. The researcher approached
the eligible participants and gave them explanations related to the study. They were
also given time for asking questions and making decisions as to whether or not to
participate in the program. Contact addresses and telephone numbers were also
Subjects were assured as to the confidentiality of their personal information. A
coding number was used to identify each subject. Only the researcher and research
assistants were able to access the data. When the study was completed the data was
destroyed and the reporting of all data would then be presented as a group.
The pregnant women were provided with the researcher and research
assistant’s name, office, and mobile phone number so that they could request
assistance if they had problems such as dizziness and headache during yoga practice.
Subjects in the experimental group were advised to take certain precautions. For
example, it was important for them to check their health status with a midwife or
obstetrician before starting yoga practice, and to avoid poses that over-stretched the
muscles, because this might cause strain, tension or pain. Pregnant women were
advised to observe carefully any signals to slow down or modify poses during practice.
They were told to stop practicing and call a midwife or physician immediately if at any
time they began to bleed. They were expected to record the problems or complications
from the yoga practice in a field note for further analysis. Throughout the study, no
subjects in the experimental group encountered any of the anticipated problems.
Moreover, no adverse effects were found in the fetus or newborn in relation to the
precautions given about thermoregulation.
Threats to Internal Validity
Internal validity is defined as the observed effect of the dependent variable that
is actually due to the action of the independent variable, and not due to the extraneous
variables (Cook & Campbell, 1979). Randomization tends to produce comparable
study groups with respect to known and unknown risk factors, and removes
investigator bias in the allocation of participants, and guarantees that statistical tests
will have valid significance levels (Friedman, Furberg, & Demets, 1998). This section
describes the potential threats to internal validity in this study and the strategies used
for controlling them.
History refers to events occurring during the course of the study (between the
pretest and the posttest), between and within the experimental and the control group
that might have influenced the final outcomes of the research. Events such as the
possible changes in nursing care, procedures, or treatment in the hospitals might have
There were two settings at public hospitals used in this study. They were
therefore organized in accord with public sector provisions required by the Thai
Ministry of Public Health. These hospitals have similar standards of care. If the
standard care, procedures, or treatments in the hospitals were changed, the effects of
the changes were the same for all the treatment carried out at these hospitals.
In order to minimize influences between or within the experimental and control
groups in the different settings and from the beginning to the end of the study, the
manipulation of core practicing of the intervention and data collecting, the orientation
given to the research assistants, and the randomization of assigning the subjects into
either the experimental or the control group were followed. Each pregnant woman
experienced the same global pattern because of this history effect.
Maturation was defined as a threat when an observed effect might be due to
such influences as the subjects being older, wiser, stronger, more experienced, and
more fatigued between the pretest and posttest, and when this maturation is not the
focus of the study. Furthermore, the pregnant state involves change, which occurs in
the individual as a function of time in the natural process with little or no nursing or
medical intervention. Therefore it was assumed that the randomization procedure at the
starting point of 26th-28th week gestation guaranteed that this threat was equally
distributed among subjects in both groups.
Testing was defined as the learning that resulted from being tested at pretest
that affects responses to the test at posttest, regardless of the introduction of the
treatment variable. In the current study, there were six timed sessions of treatment
procedure during pregnancy. One item of VASTC was used for a two-week interval
from Time 1 to Time 5, and a one-week interval from Time5 to Time 6, without the
subjects seeing the previous score. Along with the VASTC, MCDP for four-week
interval measurement was used to get more understanding of holistic comfort. Thus the
long period of time might prevent the subject from remembering the previous answers.
Moreover, the pregnant women were randomly assigned to either the experimental
group or the control group; each participant experienced the same testing conditions.
Instrumentation was a threat when an effect might be due to a change in the
measuring instrument between an initial point of data collection and a subsequent
point, and not to the treatment’s differential impact at each time interval. Changes that
occurred in the recording of the observers and raters, who could potentially make
changes in the measurements of the dependent variables, might affect internal validity.
In this study the subjects were randomly assigned to either the experimental group or
control group; the same instruments, questionnaire and visual analogue scale were
used for each pregnant woman at each time point of the data collection. Clear
guidelines for using POBS were also provided during the delivery period. Therefore,
this threat was minimized in this study.
Statistical regression was the movement of a score to the mean of a
distribution. In this study, subjects were randomly assigned to either group with an
equal probability of being assigned to the different groups of the study. Subjects were
not selected on the basis of high or low scores of dependent variables.
Selection bias was a threat, since the selection of subjects on a nonrandom
basis might produce differences in the experimental and control group. Using
randomization removed investigator bias in the allocation of participants in this study.
Interaction with selection, such as selection-history, selection-maturation, or
selection-instrumentation could cause spurious treatment effects. In the study under
review, these threats were prevented by the randomization techniques used to assign
subjects to the experimental or control group, which was assumed to guarantee equal
distribution of subjects to the groups. This would make these biases unlikely.
Attrition was the nonequivalent loss of subjects between the experimental and
control group. In recruiting participants for the study, both groups were provided with
clear pre-experiment details of the requirements and importance of the work. Hence,
they were informed in positive terms of the objectives of the study, and they were told
that, because of the requirements of the research, they should only volunteer if they
were willing to participate in either of the groups. They were told they would be
assigned randomly to the experimental or the control group, and the importance of
random assignment was briefly explained.
Some strategies were designed to diminish barriers to fully participating in the
study. These included: monetary reimbursement to research participants for their time
or for transport expenses (200 Baht per case). A small gift for the baby after birth was
also offered to subjects who volunteered to participate in the program. Furthermore,
flexible times of appointment for each session were set. These were timed to coincide
with the date for follow-ups for antenatal routine care, or at a convenient time as close
as possible to the scheduled time of the yoga program.
To minimize the rate of attrition, the researcher kept in touch with the control
group by spending around 30 minutes on a regular basis during each scheduled
appointment. The eligible number of participants in this study was a total of 74, which
remained the same until they finished the six sessions of the program. However, during
the delivery period the number dropped to 66 subjects because of the exclusion
criteria. The attrition rate in both groups of the current study was 10.81%.
Diffusion or imitation was the sharing of ideas about the treatment involved
and the program by subjects with each other and between both groups. Thus
respondents in one treatment group may acquire information intended for others. In
this study, this bias was minimal because the researcher used strategies to reduce this
threat, as stated earlier. Another possible remedy for such diffusion of knowledge of
treatment is blinding, which in this study could be only be due to a single blind of
masked assessors due to the nature of the intervention. Consequently, the participants
and the hospital staff were only informed that there were two groups in this study: the
experimental and the control group. No nurses or subjects from the antenatal clinic or
in the labor room discussed treatment with subjects in a different group. The researcher
was aware of these issues and vigilant in monitoring both participants and nurses
during the course of the study. The researcher also asked the nurses not to discuss the
study procedure with other pregnant women. It was assumed that standard care was
given equally to both groups by the hospital staff, and they supported the protocol for
the group to which each participant was assigned.
Compensatory equalization of treatment was the conscious effort to
compensate for less desirable care. This could mean nurses giving better standard care
to the control group than to the experimental group. In this study, though nurses,
physicians, or hospital administrators might perceive that the experimental group was
getting patently superior treatment to standard care, as the researcher knew, none of
them consciously interfered with the study. The researcher also made it clear to them
that the definite advantage of the treatment was still uncertain. Asking nurses not to
discuss the study procedure with other patients was another way to prevent this threat,
and it was also important to be vigilant in monitoring both physicians and nurses
during the course of the study.
Compensatory rivalry and resentful demoralization by respondents receiving
less desirable treatments could be an issue. The situation of compensatory rivalry is
one in which subjects in the control group might be motivated to attempt to reduce or
reverse the expected differences brought about by the experimental treatment. The
resentful demoralization of subjects in the control group is shown by way of reduced
effort compared with the other group. Both of these threats could have artificially
affected the results and wiped out the expected differences. In this study the subjects in
the experimental group scored higher on both VASTC and MCDP at Time 1 than did
subjects in the control group, and there were no significant differences in the relative
changes in maternal comfort during pregnancy. Thus, subjects in the control group
were not expected to compete with the experimental group, and pregnant women in the
control group were not likely to be demoralized when compared with the experimental
group. However, those in the control group knew that they were not getting an extra
intervention. The researcher was aware of these issues.
In order to decrease threats to internal validity, the instruments were composed
of instructions and a brief explanation of the data record (Roberts, Anthony, Madigan,
& Chen, 1997). Research assistants checked all questionnaires. If data were missing
unintentionally, women were asked to add more data to complete their submissions.
Each day the researcher checked the data collection instruments for accuracy, and
coded all data. Metric calculations were used with the same ruler for the VASTC and
VASPS in millimeters. In the delivery period, the subjects were told to be sure to place
marks on the lines in both of these scales in order to prevent mistakes.
During data entry, the researcher ‘cleaned’ them to decrease the errors. The
descriptive statistics were assessed to note the characteristics of the data and the
missing data to ensure the validity of the research findings. The SPSS program
(version 13.5) was used to process the data. The means, standard deviations,
frequencies, and percentages were computed to describe the characteristics of the
samples in terms of demographic characteristics and other personal data. An
independent t-test for the interval levels, and a chi-square analysis for nominal data,
was used to determine group differences.
Assumption Testing of the Statistical Analyses
First, the underlying assumptions for the t-test were examined. The dependent
variables and the control variable of trait anxiety were treated as continuous data. The
measurement of VASTC over six sessions and MCDP over three time points during
pregnancy, and VASPS, VASCT and PBOS across three time points during the active
phase of labor, together with the using of MCDL two hours after birth, were examined
for normality. No evidence of skewness, kurtosis, or univariate outliers for these
variables was found. Levene’s test for equality of variances showed homogeneity of
variances between groups. Therefore, the underlying assumptions for the t-test were
Second, the underlying assumptions for repeated measures of ANOVA were
tested. The statistic was structured with the between subjects factor for the
experimental group and the control group. Within-subjects factors are: 6 time levels of
maternal comfort by using visual analogue scale over the pregnancy period; 3 time
levels of maternal comfort by using maternal comfort questionnaires at 26th-28th,
34th, and 37th week; and 3 time levels of maternal comfort, labor pain, and behavioral
observational pain during labor. The dependent variables of maternal comfort during
pregnancy and during labor, as well as the dependent variables of labor pain and
behavioral observational pain, were normally distributed. The within-subject
correlation from the repeated measures was modeled using a compound symmetric
covariance structure. In this study, the results showed that the variances of the
differences between levels were significantly different. In order to produce a valid F-
ratio, the estimates of sphericity used to correct the degrees of freedom were Huynh-
Feldt correction when ε > 0.75 or the Greenhouse-Geisser correction when ε < 0.75
(Field, 2005). Furthermore, the homogeneity of covariance assumption, which is
inherent to analysis of covariance, was tested by constructing the interaction terms of
time by group to assess differential time means between groups. The results revealed
that the interaction between the independent variable of treatment and the covariance
were not significant. This meant that the slope of the regression line in each of the cells
is similar (Bryman & Cramer, 2001).
Third, the underlying assumptions for regression analysis were tested. In the
study, regression analysis was aimed at testing whether a relationship between the
outcome (mean difference of 1st and 6thVASTC) and explanatory variable (frequency
and duration of yoga practice) existed for the yoga group. The mean of the regression-
standardized residual based on a histogram was apparently zero. Examination of
residual scatterplots shows no violation of the assumption of homoscedasticity. The
standardized predicted values and standardized residuals had a mean of zero and amost
zero with a standard deviation of approximately or equal to 1. This was as expected if
they were normally distributed, and congruence with the P-P plot of studentized
deleted residual showed a normal distribution. No autocorrelated residuals were found
based on the Dubin-Watson value (1.447). The relation between tolerance value of
0.894 and VIF value of 1.119 confirmed their lack of multicollinearilty. In terms of
outliers, the Cook’s distance was 0.174, similar to the accepted leverage value of
0.195, indicating that there were no influential outliers in the model.
Statistical Analyses of Hypotheses Testing
The hypothesis testing in the current study and the statistics used with alpha
0.05 and a two-tailed-test were as follows.
Hypothesis 1: “The mean score of labor pain of the experimental group is lower
than that of the control group.”
To determine whether the experimental group had less pain over three time
points, repeated measures of ANOVA of VASPS and PBOS were used.
Hypothesis 2: “The mean score of maternal comfort during labor of the
experimental group is higher than that of the control group.”
To determine whether the experimental group had more comfort over three
time points, repeated measures of ANOVA of VASTC and an independent t-test of
MCDL were used.
Hypothesis 3: “The mean score of birth outcomes (length of labor, Apgar
score) of the experimental group is better than that of the control group.”
Sub-hypothesis 3.1: “The length of labor of the experimental group is shorter
than that of the control group.”
Sub-hypothesis 3.2: “Newborns in the experimental group are not at higher risk
than the newborns in the control group.”
To determine whether the experimental group had better scores of birth
outcomes; (length of labor and Apgar score at 1st and 5th minute), an independent t-
test was conducted.
Hypothesis 4: “The mean score of maternal comfort during pregnancy of the
experimental group is higher than that of the control group.”
To determine whether the experimental group had more comfort over six time
points assessed by VASTC and three time points assessed by MCQ, repeated measures
of ANOVA were used.
Hypothesis 5: “Pregnant women who undertake a greater quantity of yoga
practice (more frequently and for a longer period of time in minutes) have higher
maternal comfort than those who undertake a lesser quantity of yoga practice.”
To determine whether participants who had more quantity of yoga practice
during pregnancy had higher maternal comfort than those who had less yoga practice,
regression analysis was undertaken.
A randomized controlled trial was conducted to evaluate the effect of the yoga
program on maternal comfort, labor pain, and birth outcomes in primiparous women.
Randomized minimization techniques were used controlling for the five potential
confounding variables of maternal age, education, income, marital status, and trait
anxiety. The experimental group received a series of six 60-minute practice sessions of
the yoga program at the 26-28th, 30th, 32nd, 34th, 36th, and 37th week of gestation
respectively, whereas the control group received standard care. Repeated measures of
ANOVA (split-plot design) were used to test the differences in labor pain and maternal
comfort during labor. To determine whether the experimental group had better scores
of birth outcomes in length of labor and Apgar score, an independent t-test was used.
During pregnancy, repeated measures of ANOVA (split-plot design) were used to
investigate the differences in maternal comfort during pregnancy and regression
analysis was used to determine the effectiveness of the yoga program.